1. Field of the Invention
The present invention relates to the field of semiconductor integrated circuit (IC) manufacturing, and more specifically, to improving process control of critical dimension (CD) in lithography.
2. Discussion of Related Art
Photolithography is the process by which a pattern of features on a mask is transferred into a layer of photoresist on a wafer. A latent image is created in the photoresist by exposure to light energy in an imaging tool. Subsequently, a develop process selectively removes portions of the photoresist layer which correspond to the latent image. The pattern realized in the photoresist is then replicated through an etch process into the wafer.
The yield of microprocessors fabricated on a wafer is affected by the variability in CD of the printed features. CD is influenced by a variety of systematic and random factors. One type of effect derives from the interaction of photolithography with topography, film thickness, and reflectivity of the wafer. A second type of effect comes from non-uniformity of feature CDs on the mask. A third type of effect involves imprecision and inaccuracy of the imaging tool in aligning, leveling, focusing, or exposing the wafer. A fourth type of effect results from aberrations in the optics of the imaging tool.
A layer is often printed on a wafer with a step-and-scan imaging tool or a step-and-repeat imaging tool so as to obtain images with sufficiently high fidelity and accurate placement. A wafer is partitioned into identical small areas called fields that are sequentially exposed through the mask on the imaging tool.
Dose is the amount of light energy per unit area delivered to the wafer plane. Partial coherence, also referred to as sigma, is the numerical aperture (NA) of the illumination optics divided by the NA of the projection optics. NA is a measure of the divergence angle of light energy. NA may be varied by changing the size of an aperture stop at a pupil plane of the condenser or relay lens system.
A pattern being printed may include isolated features and nested features. In a situation where the feature is a line 110, as shown in FIG. 1 (a), the CD of the feature would be the width 114. As shown in FIG. 1 (a), an isolated feature 110 is separated from an adjacent feature 120 by a space 112 that is large compared to the CD 114 of the feature.
As shown in FIG. 1 (b), a nested feature 210 is part of a pattern 230 of repeating features and spaces that are close together. The pitch 216 of the pattern 230 is the sum of the CD 214 of a feature 210 and the space 212 between the feature 210 and a neighboring feature 220. In another situation, the feature may be a round contact hole 310, as shown in FIG. 1 (c). In that case, the CD would be the diameter 314.
As the CDs of the features on a microprocessor become smaller, the within-field CD variability consumes an increasingly larger portion of the overall CD error budget. The within-field CD variability encompasses isolated-dense (iso-dense) bias and horizontal-vertical (H-V) bias. Iso-dense bias involves CD variability related to proximity to other features. H-V bias involves CD variability related to aberrations in the optics or related to non-uniformities in partial coherence between the horizontal direction and the vertical direction.
As the critical dimension (CD) of a printed feature approaches the wavelength of the exposure light, interference of the light diffracted by the features will tend to degrade pattern fidelity. Then it becomes necessary to use a resolution enhancement technique (RET), such as pupil filtering (PF), phase-shifting mask (PSM), or off-axis illumination (OAI). However, application of a RET is often difficult.
PF improves the depth-of-focus (DOF) by putting an intensity filter or a phase filter into the projection lens pupil of an imaging tool. However, accessing the pupil planes in an exposure tool may risk contaminating or damaging the optics within the imaging tool.
PSM improves resolution and DOF by introducing a 180-degree difference in phase between the light transmitted through adjacent openings to cause destructive interference at the boundary. However, the design of a PSM is very complicated due to phase conflicts and artifacts. A PSM is also difficult to fabricate, inspect, and repair due to the three-dimensional structure required to form the phase shifter.
Conventional illumination is usually used to print a feature that has a CD that is large relative to the exposure wavelength of the imaging tool. FIG. 2 (a) shows a conventional aperture 410 that may be used to produce conventional illumination. The conventional aperture 410 has a circular opening 415 in an opaque plate 412.
As the pitch of nested features shrinks, it is often desirable to block out a large fraction of the exposure light that is not useful in printing the features. Such techniques are called OAI. OAI uses a centrally-obstructed aperture to illuminate a mask with the obliquely incident components of the exposure light. Examples of OAI include annular illumination, dipole illumination, and quadrupole illumination. However, conventional implementations of OAI improve the resolution and depth of focus of nested patterns at the expense of a larger iso-dense bias.
FIG. 2 (b) shows an annular aperture 510 that may be used to produce OAI. The annular aperture 510 has a ring-shaped opening 515 in an opaque plate 512. Alternatively, the annulus 515 may be considered to be a circular opening, equivalent in size to the annulus 515, that has a central disk 520 that is opaque.
FIG. 2 (c) shows a quadrupole aperture 610 that may also be used to produce OAI. The quadrupole aperture 610 may be considered as having an imaginary square 617 within an opaque plate 612 that has a circular opening 615 located at each corner of the imaginary square 617.